Search

Background Malnutrition underlies 45% of under-5 deaths globally. Severe acute malnutrition (SAM) is the most serious form of undernutrition, characterized by wasting with or without edema. Mortality remains high (10%–40%) among children requiring hospitalization for complicated SAM. Objectives We aimed to systematically document the factors independently associated with inpatient mortality in children with SAM. Methods Embase, Ovid MEDINE, the Cochrane Library, and clinicaltrials.gov were searched for articles published between January 2000 and January 2020, using a prespecified protocol. Eligible studies included children aged ≤59 mo hospitalized with SAM and used multivariable analysis to assess the baseline factors independently associated with inpatient mortality. Random-effects meta-analysis, stratified by the stated measure of effect, was used where >20% of studies included the same factor in analyses. Results Twenty-eight of 1432 studies fulfilled inclusion criteria: 19 studies included all children with SAM and 9 included specific subgroups of children with SAM. All 19 main studies were from 8 countries across Africa, with a median of 400 children/study. The mean inpatient mortality was 15.7% (95% CI: 10.4%, 21.0%) and HIV prevalence ranged from 2.1% to 51%. Nine factors were included in the meta-analysis, stratified by HR and OR. HIV infection (HR: 4.32; 95% CI: 2.31, 8.08), weight-for-height z score (WHZ) (OR: 0.44; 95% CI: 0.24, 0.80), diarrhea (HR: 2.84; 95% CI: 1.40, 5.75), pneumonia (HR: 1.89; 95% CI: 1.19, 3.02), presence of shock (HR: 3.67; 95% CI: 2.24, 6.03), and lack of appetite (HR: 2.16; 95% CI: 1.48, 3.16) were associated with increased mortality, whereas child age and sex were not. The association between edema and mortality was difficult to ascertain from the available studies. Conclusions HIV infection, diarrhea, pneumonia, shock, lack of appetite, and lower WHZ are independent predictors of inpatient mortality in children with SAM. These factors may help to risk-stratify children being hospitalized with complicated SAM. This systematic review/meta-analysis protocol was registered at www.crd.york.ac.uk/prospero as CRD42019152267.